Communication NBB_2021_24 / Evaluation of technical provisions for individual health insurance

Boulevard de Berlaimont 14- BE-1000 Brussels tel. +32 2 221 38 12 -fax +32 2 221 31 04 company number: 0203.201.340 RPM Brussels www.nbb.be NATIONAL BANK OF BELGIUM DE BELGISCHE NATIONALBANK Eurosystem Communication Brussels, 21 December 2021 Reference: NBB_2021_24 Your contact:
Frank Van Steen and Stany Zabinski tel. +32 2 221 21 23/3467 frank.vansteen@nbb.be / stany.zabinski@nbb.be Evaluation of technical provisions for individual health insurance Scope Insurance or reinsurance undertakings governed by Belgian law that are subject to the Act of 13 March 2016 on the status and supervision of insurance or reinsurance undertakings. Summary/Objective This communication describes a reference scenario that the National Bank of Belgium will use as a lower limit to assess the prudence of assumptions surrounding the modelling of medical inflation and which will serve as a basis for identifying and quantifying structural imbalances in the pricing of health insurance. Structure

1. Context
2. Verification and entry into force
3. Methodology of the reference scenario
4. Description of the reference scenario
   NBB_2021_24 - 21 December 2021 Communication - Bus. 1/6

Dear Sir or Madam,

1. Context
   The Verwilghen Act provides specific provisions for individual health insurance. Thus, a health insurer is required to offer lifetime contracts, and the possibilities to modify the technical bases of premiums and coverage conditions are legally limited. Under the legal framework, policies can only be adapted in the following cases1:
   1) by mutual agreement and in the interest of the insured;
   2) in the event of a change in the insured's personal situation (profession and social status);
   3) based on the consumer price index or a specific index (the medical index); or
   4) by an intervention of the National Bank of Belgium (hereinafter "the Bank").
   With the entry into force of the Act of 13 March 2016 on the status and supervision of insurance or reinsurance undertakings (the "Insurance Supervision Act"), the pricing adaptation referred to in point 4 above has become a specific prerogative of the Bank and falls within the framework of possible recovery measures available to the Bank. Health insurers offering individual health insurance policies face complex and sensitive choices in terms of valuation under Solvency II. Thus, the very long projection horizon makes the valuation of the corresponding technical provisions particularly sensitive to the underlying model parameters. It is also necessary to make adequate projections of the evolution of medical costs over these potentially (very) distant maturities. Objectivity, stability and prudence are therefore the key words for the valuation of the best estimate for individual health insurance policies. To examine the sensitivities surrounding the valuation of the best estimate and the mutual level playing field, the Bank decided to conduct a broad horizontal analysis focused on the modelling of assumptions regarding the evolution of medical inflation. The results confirmed, in addition to market heterogeneity, the enormous sensitivity of the best estimate valuation to these assumptions and, consequently, the need for clear guidelines for the modelling of this type of product. Based on the results of the analysis, the Bank decided to develop a scenario that will be used as a lower limit to assess the prudence of the medical inflation modelling assumptions used for the calculation of technical provisions. Based on the principle of proportionality, the insurance undertaking may deviate from this provided that it does not result in an underestimation of technical provisions. The Bank will also use this scenario as a basis for quantifying potential pricing imbalances.

2. Verification and entry into force
   Pursuant to Article 272 of Delegated Regulation (EU) 2015/35, the actuarial function must ensure that the most appropriate approximations for the purpose of calculating the best estimate are used and, in the case of individual health insurance, this amounts to verifying that any deviations from the reference scenario do not lead to an underestimation of technical provisions. This verification must be carried out annually. The communication applies for the first time to figures at the reference date of 31/12/2021. 1 See Article 204 of the Act of 4 April 2014 on insurance. Communication - Bus. 2/6 NBB_2021_24 - 21 December 2021

3. Methodology of the reference scenario
   The reference scenario is based on consumer price inflation increased by a wedge. Indeed, medical costs should grow more vigorously than the HICP since they follow prosperity and are therefore at least in phase with real growth. However, they should continue their short-term growth due to technological innovation, which continues to push them upwards. The wedge therefore represents the sum of real growth increased by this technological innovation. As the best estimate valuation is based on market-consistent principles, it is appropriate in this case to use as much information as possible from liquid markets. For HICP increases, there are liquid inflation-indexed swaps (inflation swaps) that provide an estimate at the euro area level. At the national level, there are sometimes also inflation-linked bonds, which, however, generally present higher HICP expectations due to illiquidity premiums contained in the valuation. These instruments therefore seem less appropriate. For the calibration of the wedge, there are no financial instruments: historical calibration must be used. GDP-linked bonds are, however, already available, but they are still in the design stage. For the long term, the inflation curve is inspired by the risk-free interest rate curve as produced by EIOPA. In this case too, a Smith-Wilson extrapolation will be used from a maturity of 20 to 60 years up to an Ultimate Forward Medical Inflation Rate (hereinafter the "UFMIR"). The coherence of the extrapolation is important to ensure that no additional volatility results from inconsistency between the risk-free interest rate curve and medical inflation curves. In addition, the UFMIR is based on the long-term expectations used by the Study Committee on Ageing2 and the European Commission3 for the projection of public health expenditure. Indeed, an income elasticity of 1.2 to 1.5 was observed in the years 1960 to 1980 during different years and in different countries. This finding is, however, explained by the establishment of social security after the Second World War and did not constitute a long-term effect. Income elasticity has recently converged to 1. This convergence indicates that in the long term, the real increase in public health expenditure is in line with the real growth of GDP per capita, which corresponds to 1.5%. In addition, it is assumed that the growth of medical costs for the insurance sector is in line with the growth of public expenditure. Indeed, a stability of policy is assumed in the projection of the best estimate. Moreover, it is assumed that HICP growth is in line with the 2% target of the European Central Bank's monetary policy. The long-term nominal growth of private medical costs therefore amounts to 3.5% (= 1.5% x 1 + 2%).

4. Description of the reference scenario
   4.1. General modalities
   The reference scenario relates exclusively to assumptions regarding the expected evolution of the inflation of claims of undertakings and the expected evolution of general medical inflation at the market level for individual health insurance such as hospitalization, outpatient care and dental care. It combines the use of observed developments in premiums and claim costs in the short term with long-term macroeconomic elements. Concretely, the scenario has the following characteristics regarding the modelling of the evolution of medical inflations:

• short-term evolution of medical inflations: based on a European inflation curve increased by an inflation wedge.
  o European inflation curve: constructed using zero-coupon inflation indexed swaps (ZCIIS) based on the Harmonised Index of Consumer Prices excluding Tobacco (HICPxT) of the euro area. o Inflation wedge: differential between European inflation and Belgian claim inflation.
  " specific medical inflation (important for the evolution of claim payments): calibration based on a sufficiently large set of historical observations specific to the undertaking. The wedge is defined as the average differential between specific claim inflation and HICPxT inflation of the euro area.
  The medical inflation curve for claims does not apply to indemnity-based insurance, but only to insurance covering expenses. The projection for indemnity-based insurance is not linked to medical inflation.
  " general medical inflation (at market level, therefore important for the evolution of premium income via the medical index): based on historical observations of the base indices of Article 7 of the Royal Decree of 18 March 2016 determining specific indices, possibly supplemented by expert judgement if the available data are not deemed sufficiently significant. The wedge is defined as the average differential between general claim inflation and HICPxT inflation of the euro area. Given the volatility of recent observations, it is requested to limit oneself to the overall medical index without taking into account age categories. By analogy with the "liquid" part of the EIOPA risk-free interest rate curve, the short term would reach a Last Liquid Point of 20 years.
• long-term evolution of medical inflation: forecasts of medium and long-term medical inflation can be very diverse and are surrounded by a very high degree of uncertainty. That said, this uncertainty does not prejudice the obligation to formulate well-founded and objective hypotheses for the long term as well. For the long-term part, macroeconomic principles are therefore taken into account, which results in a long-term medical inflation that is equal, according to the following formula, to an ultimate forward medical inflation rate of 3.5%:
  UFMIR = income growth (1.5%) x elasticity (1) + base inflation (2%) = 3.5%. where the long-term estimate of GDP growth by the Federal Planning Bureau (1.5%) and the inflation target of the European Central Bank (2%) are taken into account. By setting income elasticity to 1, it is assumed that the evolution of medical costs, on a real per capita basis, follows general income growth.
• The Smith-Wilson extrapolation method is prescribed for the convergence of short-term results to the ultimate forward medical inflation rate prescribed above. As mentioned above, this method is already well known in the context of constructing the EIOPA risk-free interest rate curve. It will therefore be used to allow the curve to converge, from the Last Liquid Point of 20 years, to 3.5% at a maturity of 60 years. The technical specifications of this method can be consulted on the EIOPA website4.
  https://www.eiopa.europa.eu/tools-and-data/risk-free-interest-rate-term-structures_en - Risk-free rates previous releases and preparatory phase.
  Communication - Bus. 4/6 NBB_2021_24 - 21 December 2021

4.2. Particular points of attention

• Point of attention 1: Calibration of wedges - Historical data
  Specific medical inflation wedge: Precision prevails here. If a large history of detailed and reliable data is available, a history as long as possible should be used.
  General medical inflation wedge: Stability prevails here. The number of medical indices calculated and published by the Federal Public Service Economy in accordance with the Royal Decree of 18 March 2016 is quite limited. Where possible, these data should be supplemented by older relevant statistics that are also available, for example the statistics published by Assuralia. It is important to approach these data with caution (cf. point 3).

• Point of attention 2: Calibration of wedges - Data granularity
  Specific medical inflation wedge: Without prejudice to the principle of proportionality, claim profiles must be established by product and sex, the granularity of products being in line with that used for the medical index.
  General medical inflation wedge: Product-level estimation is sufficient in this case, the granularity of products being in line with that used for the medical index.

• Point of attention 3: Calibration of wedges - Representativeness of data
  If, due to changes in regulation and legislation or developments in underwriting policy and product assortment, the evolution of medical costs in the past is not representative of the future, this should be remedied. If, moreover, the data used relating to medical inflation only represent part of the undertakings subject to the medical index, these data must be adapted so that they are representative of the entire market. Finally, expert judgements can also be applied to data relating to health crises that have a strong short-term impact, but whose long-term impact is considered less significant.

• Point of attention 4: Calculation of the wedge
  The calculation of the wedge must be carried out using one of the following methods:
  1) The wedge can be calculated on the one hand as an average differential between specific claim inflation and HICPxT inflation of the euro area.
  2) In that case, the undertaking may also use a specific model to establish a rate curve for the wedge.

• Point of attention 5: Construction of the inflation curve
  Since convergence depends on the short-term curve, this means that the convergence of the Smith-Wilson methodology must be calibrated to ensure that the UFMIR is reached at 60 years. In the event of inadequate calibration, it would be reached earlier or later.
  NBB_2021_24 - 21 December 2021 Communication - Bus. 5/6

• Point of attention 6: Coherence with the Verwilghen Act
  The inflation curve based on general inflation is an estimate of future base indices. It can therefore, in accordance with Article 7 of the Royal Decree of 18 March 2016, be used to construct future specific indices for the calculation of the evolution of future premium income.

• Point 7: Possibility of stochastic valuation
  The communication presents the modalities for a deterministic calculation. If the undertaking has future management activity that depends on the specific value of the HICP relative to the medical index, the undertaking may, for the calculation of the Best Estimate, deviate from this deterministic calculation and use a stochastic model for the HICP, the wedge and the medical index. The undertaking applies the management activity for each individual simulation based on the concrete value of the HICP, the wedge and the medical index at the time in the simulation. The verification by the actuarial function that any deviations from the reference scenario do not lead to an underestimation of technical provisions, in the case of a stochastic valuation model, must be done based on the central Best Estimate, in line with a single central scenario. The final Best Estimate, after verification that technical provisions are not underestimated relative to the reference scenario, is thus equal to:
  BE > max(BE_ref, BE_central) + TVOG
  With:

• BE_ref the deterministic Best Estimate based on the medical inflation from the reference scenario;

• BE_central the central Best Estimate according to the undertaking's own medical inflation hypothesis based on a single central scenario;

• TVOG the time value of options and financial guarantees calculated as the difference between the stochastic Best Estimate and the central Best Estimate BE_stoch - BE_central

• BE_stoch the stochastic Best Estimate according to the undertaking's own medical inflation hypothesis based on a complete set of stochastic scenarios,
  A copy of this communication is sent to the approved commissioner(s) of your undertaking.
  Sincere regards.
  Pierre Wunsch
  Governor
  Communication - Bus. 6/6 NBB_2021_24 - 21 December 2021